How Square Capital uses traditional and non-traditional data sources to extend loans to merchants

Square was able to democratize small business lending by using traditional banking methods as well as using machine learning algorithms to better underwrite small businesses .

Square is a financial services player mostly known for its merchant services to small businesses[1]. The company offered e-readers and other software services that would allow small businesses to accept card payments. By 2014, Square launched Square Capital which would be the unit that would offer small business financing to merchants who would already be using Square.

 

The original product Square Capital was called “merchant cash advances,” which gave merchants access to funds upfront, and the loan provider would then take the principal and loan payments from merchants’ sales. This model was seen to be good for the small business as the payments would adjust based on the business’s sales, and the charges were taken automatically, which reduced the risk of the loan provider. Merchant cash advances, however, were not a proprietary product themselves; it is the way that Square accessed and underwrote the risk that made this an exciting model.

 

Square today looks at over 400 data points before making a loan decision and does this whole process automatically. Businesses who apply for loans will usually know within minutes if they are accepted for their loan applications. This is distinctive from the traditional model in which banks would make loans to businesses. The standard model would have small companies going to their local business branch and applying through a branch employee. The process could either take weeks or even months to complete. It was often unclear what information would be required from the loan applicant, and there was a bias on the relationship the loan officers would have with the applicant. Standard data included bank statements, receipts, invoices and long term contracts that the business may have. 

 

Square’s model looks at the traditional data that banks would look at and now also use machine learning algorithms to access other risk factors. Traditional factors that Square looks at include transaction history, revenue volume and growth, the mix of customers and churn. Banks have similar data, but Square set up its systems to analyze the data better. Square also looked at non-traditional factors discovered through the algorithms and not necessarily through traditional banking methods. One of the factors that have proven to be a good predictor of default is whether a business has the business name in their email address, for example.

 

This process generates value as Square is currently looking at a market that was previously overlooked. The market is loans under $100K for small businesses. Banks do not like these size loans because there is a high rate of fixed cost to processing a loan using traditional banking method, so the loans with ticket sizes of $100K or less become less valuable for the banks to process. What was also found was that Square Capital was also disbursing loans to a larger share of female and minority entrepreneurs[2]. A possible explanation for this was because Square Capital processed the complete loan application using their algorithms; no bias of gender and race could interfere with the loan process.

The most significant asset that Square has hence is not the hardware, including the Square devices or the software that powers their systems. The most critical asset that Square has is the data it has accumulated from its merchant customers. 

 

Square’s issue in becoming a significant player in small business lending was to get enough data on customers to start giving out these loans. Square thought about this to first become a leading payment gateway for small businesses struggling to accept card payments. In that way, they were able to digitize small business information and then run analysis to determine companies’ creditworthiness. The next challenge that Square will face is with data privacy and the rise of Open Banking type initiatives like in Europe. Consumers and businesses have more freedom to decide what data they should share with companies, and businesses are forced only to use consumer data once given permission. Square merchants would hence ruin the model if businesses wanted Square to stop access their data. The more significant threat of Open Banking is that other companies would be able to access the data Square has accumulated on their clients to give them products, and this would be based on the clients’ consent. This would be a problematic situation for Square as they would not have the information advantage it currently has on other lenders.

 

 

  [1] https://en.wikipedia.org/wiki/Square,_Inc.

[2] https://www.americanbanker.com/news/how-square-turns-unconventional-credit-data-into-loans-for-tiny-businesses

Previous:

Tala – How data boosts access to credit for low-income individuals in emerging economies

Next:

Symrise: AI to create new perfumes

Student comments on How Square Capital uses traditional and non-traditional data sources to extend loans to merchants

  1. Hi Molefe — great article! It is really interesting to read how Square has built out their loans business and how ML has played a significant role in helping them manage risk.

    The potential world where Square is no longer able to obtain access to the necessary data to assess loan applications seems like a tremendous risk to this part of the business! Do you see a path forward for Square in this line of business if they are no longer able to access necessary data?

  2. Great article Molefe! My biggest question, and one I’m hoping to answer through my own startup, is what happens when Square encounters a business owner without a FICO score, especially when we’re talking about micro-entrepreneurs? There are millions of professionals that don’t have SSNs in the US today, but with a ton of potential for growth. I don’t know if you have the answer, but this is a question that is constantly on my mind. Is it possible to serve this community with enough alternative data without having the one metric that everybody uses: the FICO score?

  3. Thanks for the insight Molefe! I love how Square has used data in a way that is indirectly helping small businesses succeed and has helped reduce some of the biases built into the financial system. Building on Bruno’s point, it would be great if Square could continue building up the data necessary to start providing credit in the informal economies of the world. Do you think this could happen?

Leave a comment